CHAPTER 32. Advanced Metastatic Disease. Randomized Trials Versus Best Supportive Care. Chemotherapy Trials

Transcription

1 21786_Ch32.qxd_ /20/05 9:09 AM Page 447 CHAPTER 32 Advanced Metastatic Disease Marcel Rozencweig, MD Daniel D. Von Hoff, MD Adenocarcinoma of the pancreas is an aggressive and rapidly fatal disease that has shown little sensitivity to currently available anti-cancer agents. The American Cancer Society estimates for 2004 in the United States anticipate 31,860 new cases of pancreas cancer and 31,270 deaths.1 The vast majority of patients present with disease that is no longer suitable for surgical cure, and debilitating symptoms, including pain, emesis, weight loss, and weakness, are common features during the clinical course of the disease.2 Large-scale chemotherapy trials have generally yielded marginal tumor response rates, and survival has been only minimally impacted.3 Performance status and stage of disease are prominent prognostic factors in advanced disease. The median survival is 6 10 months for patients with locally advanced disease and 3 6 months for those with metastatic disease.4 This chapter reviews the randomized trials testing anti-cancer agents for locally advanced and metastatic pancreatic cancer. The review includes phase II and phase III trials. Most phase II trials were designed for screening purposes and focused on response rates based on tumor shrinkage. It has since been recognized, however, that low tumor response rates may be observed despite prolonged survival, emphasizing the need for other surrogate endpoints of efficacy. Response based on general symptom improvement has been recently proposed as a measure of efficacy for chemotherapeutic agents in pancreatic cancer. Whether this clinical benefit response will prove to be a more useful endpoint for screening new therapeutic approaches is still uncertain. Survival remains the most reliable endpoint in clinical trials for this disease, as long as study design is adequate and statistical power is commensurate with a small, but clinically significant improvement in outcome. Randomized Trials Versus Best Supportive Care Chemotherapy Trials A number of randomized trials have compared combination chemotherapy with best supportive care in patients with advanced pancreatic cancer (Table 32.1). Although the results of these studies have been inconsistent, they did provide some support for the concept that chemotherapy can be beneficial in patients with metastatic cancer of the pancreas. These studies accrued small numbers of patients. Most had less than 30 patients per treatment arm. Survival in 447

2 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease Table 32.1 Author(s) Mallinson et al. 5 Palmer et al. 6 Glimelius et al. 7 Andersen et al. 8 Frey et al. 9 Andren-Sandberg et al. 10 Randomized Trials of Chemotherapy or Hormone Therapy Versus Best Supportive Care No. of Treatment Patients Median Survival Cyclophosphamide,, Vincristine, Methotrexate, Doxorubicin, Mitomycin C, Leucovorin, Etoposide, BCNU, CCNU, CCNU, vincristine weeks 9 weeks 33 weeks 15 weeks 6 months 2.5 months 13 weeks 14 weeks 3 months 3.9 months 5 months 4 months P Value Bakkevold et al. 11 Keating et al. 12 Greenway 13 Abbreviation:, not reported. Tamoxifen Tamoxifen Cyproterone acetate Flutamide days 122 days 5.25 months 4.25 months 3.0 months 226 days 120 days vs control 0.50 vs control the control group ranged consistently between 2 and 5 months. Chemotherapy consisted of 5-fluorouracil () combinations, and in the three chemotherapy trials reporting a survival benefit, these combinations were associated with a median survival of 6 to 10 months. In these trials, was administered at 2- to 4-week intervals. The other three chemotherapy trials reported here used combination regimens of and nitrosourea derivatives in cycles repeated every 6 to 8 weeks; all failed to show any difference in outcome, perhaps, in part, as a result of suboptimal administration. The first randomized study of combination chemotherapy versus best supportive care was published by Mallinson et al. in A total of 40 patients with unresectable pancreatic cancer were randomized to receive no chemotherapy or an initiation course of, methotrexate, vincristine, and cyclophosphamide followed after 4 weeks by a maintenance program with a combination of and mitomycin C. Patients in the chemotherapy group had a median survival of 44 weeks, which was significantly longer than the median survival of 9 weeks in the group that did not receive chemotherapy (P ). The difference in survival remained statistically significant among patients with histologically confirmed (n 26) or unconfirmed diagnosis (n 14), as well as those with local tumor only (n 25) or those with hepatic metastases (n 15). The favorable outcome in the chemotherapy group in this trial might have been amplified by the unusually short median survival in the control group. The study of Palmer et al. 6 randomized 43 patients with unresectable pancreatic cancer. Histopathologic confirmation of the diagnosis was obtained in 31 of these patients. Chemotherapy consisted of the FAM regimen designed at Georgetown University. That regimen combines 600 mg/m 2 on days 1, 8, 29, and 36; doxorubicin 30 mg/m 2 on days 1 and 29; and mitomycin C 10 mg/m 2 on day 1 with cycles repeated every 8 weeks. Compared with the control group that received no chemotherapy, the median survival in the FAM group was more than twice as long (33 vs 15 weeks, P 0.002). Most chemotherapy-treated patients experienced toxicities, but these toxicities were usually mild and of short duration. There were only two instances of treatment delays because of neutropenia.

3 21786_Ch32.qxd_ /20/05 9:09 AM Page 449 Randomized Trials Versus 449 Another randomized study demonstrated that combination chemotherapy not only prolonged survival but also improved quality of life, as measured by the European Organization for Research and Treatment of Cancer Quality-of-Life Questionnaire-C30. 7 The trial randomized 93 patients with advanced pancreatic or biliary cancer to either chemotherapy in addition to best supportive care or to best supportive care alone. Chemotherapy was allowed in the latter group if satisfactory palliation could not be achieved with supportive measures. All patients had their diagnosis histologically verified. Randomization was stratified by primary tumor site (pancreatic and biliary) and participating hospital. Chemotherapy consisted of, leucovorin, and etoposide. The older population and poor performance status patients received the same chemotherapy regimen without etoposide. Among patients with pancreatic cancer, overall survival was significantly longer for those randomized to chemotherapy than for those assigned to the best supportive care only treatment group (6 and 2.5 months, respectively, P 0.05). Furthermore, 38% of patients in the chemotherapy group had a favorable quality-of-life outcome compared with only 13% of those in the control group. Three other randomized studies failed to detect a benefit in survival or in quality of life (see Table 32.1). With a combination of and carmustine, the median survival and 1-year survival were not significantly different than they were for control patients. 8 The median survival was 13 weeks in the treatment group versus 14 weeks for the control group (P 0.80), and 1-year survival was 10% in both groups. The Veterans Administration Surgical Adjuvant Cancer Chemotherapy Study Group compared the nitrosourea lomustine in combination with to best supportive care in a randomized study of 152 patients with histologically proven inoperable pancreatic cancer. 9 The control subjects were to receive no chemotherapy. Both treatment groups were comparable in terms of age, weight loss, extent of metastases, and prior surgical procedure. Toxic effects for the most part were reportedly mild. There was no survival benefit associated with drug therapy. In fact, there was a trend suggesting a decrease in survival in the drug treatment group relative to the control group, but the difference (medians of 3.0 vs 3.9 months) did not reach statistical significance (P 0.17). A triple drug regimen consisting of vincristine 1 mg/m 2 intravenously on day 1, followed by oral lomustine 40 mg/m 2 on days 2 and 3 plus oral 500 mg daily on days 2 5 was compared with the best supportive care in another randomized study involving 47 patients. 10 The median survival was 5 months in the treatment group and 4 months in the control group; that difference was not statistically significant. All patients randomized to chemotherapy in this study reported more or less marked gastrointestinal adverse events within days of each treatment cycle. Hormone Therapy Trials The concept that hormonal therapy might be useful in the treatment of pancreatic cancer is derived from the presence of estrogen and androgen receptors in pancreatic cancer cells. However, anti-estrogen treatment with tamoxifen has not proven to be an effective approach to improving survival in patients with pancreatic cancer. The median survival in a randomized trial comparing tamoxifen with placebo was 115 days for tamoxifen-treated patients compared with 122 days for patients in the placebo group. 11 Results were more encouraging in another trial where the median survival was longer for patients in the tamoxifen group than in the group receiving no active therapy (5.25 vs 3.0 months). 12 The difference was not statistically significant, but this was a small trial. The trial also included a third arm testing cyproterone acetate, an anti-androgen. No effect on survival was detected with this agent either versus the no active treatment arm. Still, positive results from a more recent double-blind placebo controlled trial of flutamide suggest a possible role for androgen receptor antagonist in pancreatic cancer. 13 Patients were treated with flutamide 250 mg three times daily or placebo. This was a very small study (n 49), and information on prognostic factors was missing. There was no apparent difference between treatment arms in terms of tumor regression, but an intent-to-treat analysis showed a significant prolongation of survival in the flutamide group (226 vs 120 days, P 0.01). Flutamide treatment was well tolerated with minimal side effects. Further randomized trials are needed to evaluate the role of hormonal approaches in this disease. Randomized Trials Versus was the reference standard to treat patients with advanced cancer of the pancreas for more than 2 decades. Its modest anti-tumor effect prompted large efforts aimed at identifying new active chemotherapeutic agents and more active combinations. Several groups used randomized trials to screen for new and promising approaches, with a focus on response rates in small numbers of patients with measurable disease. Many investigational agents were tested, mostly with disappointing results, in major research programs orchestrated by several cooperative groups, including the Gastrointestinal Tumor Study

4 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease Table 32.2 Author(s) Kovach et al. 19 Stolinsky et al. 20 Maisey et al. 21 Cullinan et al. 22 Takada et al. 23 Moertel et al. 26 Awrich et al. 28 Ducreux et al. 32 Cullinan et al. 34 Rougier et al. 35 Burch et al. 36 Randomized Trials of Based Combination Regimens Versus Alone No. of Treatment Patients Median Survival P Value, BCNU BCNU (oral) (intravenous) PVI, Mitomycin C PVI, Doxorubicin, Mitomycin C, Doxorubicin, Doxorubicin, Mitomycin C,Streptozotocin,Spironolactone, Streptozotocin, Spironolactone, Streptozotocin,Tubercidin CI, Cisplatin Mallinson Regimen, Doxorubicin, Cisplatin Oxaliplatin, Oxaliplatin Octreotide with Leucovorin * Abbreviations:, not reported; NS, not significant; PVI, prolonged venous infusion; Mallinson regimen, induction with, cyclophosphamide, methotrexate, and vincristine followed by maintenance with, mitomycin C; CI, continuous infusion. * There was no statistically significant difference between octreotide and the two arms combined (P 0.80) or between the two chemotherapy arms (P 0.08) days 110 days 5.1 months 6.5 months 6.2 months 6.0 months 16 weeks 18 weeks 15 weeks 21 weeks 112 days 102 days 4.5 months 3.5 months 3.5 months 8.5 months 3.4 months 2.4 months 0.34 NS Group, the Southwest Oncology Group, 17 and the Eastern Cooperative Oncology Group. 18 Much of this work involved unfavorable patient selections, with prior chemotherapy, far-advanced disease, and median survival times often of 8-10 weeks. has been combined with a number of agents, predominantly nitrosourea derivatives (BCNU, CCNU, MeCCNU, streptozotocin), mitomycin C, and doxorubicin. These combination regimens have not demonstrated consistent and compelling superiority over single-agent. The gains from combination therapy were modest, and there was no positive impact on survival. The risk of excess toxicity actually often diminished the feasibility of based combinations. The following section summarizes the long trail of randomized chemotherapy trials that compared a wide variety of regimens with single-agent therapy (Table 32.2). An early trial of the Mayo Clinic compared given intravenously (13.5 mg/kg/day five repeated every 5 weeks) with BCNU and the combination of these two agents. 19 Of 82 patients evaluated in that trial, 49 had a primary pancreatic lesion established at surgery, and the remainder had a presumed diagnosis of pancreatic cancer based on histology of a metastatic lesion, negative barium roentgenologic studies, and a convincing clinical presentation. The response rates achieved with alone

5 21786_Ch32.qxd_ /20/05 9:09 AM Page 451 Randomized Trials Versus 451 (16%, 5 of 31) or in combination (33%, 10 of 30) were significantly higher than with BCNU alone (0 of 21) (P 0.05), but the combination was not significantly better than alone, and there was no difference in survival among the three treatment options. A randomized phase II trial compared oral versus intravenous administration of given at 15 mg/kg/week in both arms. 20 No minimum performance status was required, but the median performance status at entry was 70 on the Karnofsky scale in both arms. Three responses were observed among 14 patients with intravenous therapy, whereas none of the 16 patients receiving oral achieved a response. The corresponding median survival times were 110 and 53 days, respectively. The potential value of mitomycin C was evaluated in combination with relative to alone. 21 Patients in both arms had protracted intravenous infusions of at a daily dose of 300 mg/m 2. Those randomized to mitomycin C treatment also received a bolus of 7 mg/m 2 of mitomycin C on day 1. Treatment was repeated at 6-week intervals until four courses had been administered. Among evaluable patients, overall response rates favored the combination over the single agent (18.5% [n 92] vs 8.6% [n 105], P 0.04). This statistically significant difference did not, however, translate into a significant difference in failure-free or overall survival. Both chemotherapy regimens were well tolerated. A three-arm randomized trial compared single-agent to a two-drug combination of plus doxorubicin (FA) and a three-drug combination of plus doxorubicin and mitomycin C (FAM). 22 Single-agent was given at a dose of 500 mg/m 2 daily for 5 days and repeated every 4 weeks for two cycles and every 5 weeks thereafter. The FA regimen consisted of given by a 4-day course at a daily dose of 400 mg/m 2 plus 40 mg/m 2 of doxorubicin on day 1 of a 4-week cycle. After two cycles, treatment was repeated every 5 weeks. The FAM regimen was comparable to the combination of, doxorubicin, and mitomycin C that had demonstrated improved survival relative to best supportive care. 6 A total of 144 patients with pancreatic cancer were randomized. The number of patients with measurable disease was too small for any meaningful comparison of response rates. However, survival curves and distribution of progression times between the three treatment arms were superimposable. Diarrhea and stomatitis occurred more frequently with single-agent treatment, but the FAM regimen produced more nausea, vomiting, and myelosuppression. A modified FAM regimen was also compared with alone in a multicenter randomized trial carried out in Japan among 81 patients with nonresectable carcinoma of the pancreas or the biliary tract. 23 That trial also failed to detect a difference in time to progression or survival. Streptozotocin is a nitrosourea derivative that underwent extensive testing in pancreatic cancer. Multiple combinations were tested, with mixed results, in several randomized trials. Bukowski et al. 24 evaluated the addition of streptozotocin to a regimen of plus mitomycin C, with given as a 24-hour intravenous infusion for 4 consecutive days every 4 weeks. The trial accrued 181 patients with inoperable pancreatic adenocarcinoma and no prior chemotherapy. Twenty percent of the patients were not evaluable and were excluded from the analysis of response and survival. Response rates were significantly higher (34% vs 8%) when streptozotocin was included in the regimen (P 0.009). The median survival, however, was virtually the same (18 and 17 weeks) with and without streptozotocin, respectively. Gastrointestinal toxicities were more frequent and more severe when streptozotocin was added. Mild and reversible renal toxicity was also associated with the use of streptozotocin. There were four treatment-related deaths in the plus mitomycin arm and one in the SMF arm. All deaths were attributed to severe myelosuppression. A different SMF combination was associated with prolonged survival in another randomized trial. 25 That trial compared the SMF regimen developed at Georgetown University to a regimen of cisplatin, cytosine arabinoside, and caffeine (CAC). SMF consisted of 8-week cycles with streptozotocin (1 g/m 2 ) and (600 mg/m 2 ) given on days 1, 8, 29, and 36 and mitomycin C (10 mg/m 2 ) given on day 1. In the CAC arm, patients received a single dose of cisplatin 100 mg/m 2 and cytosine arabinoside 2 g/m 2, given as a 3-hour infusion for two doses, 12 hours apart, followed by caffeine administration. Caffeine was included because of its putative interference with DNA repair. Cycles were repeated every 4 weeks. Eligibility criteria included a Karnofsky performance status of 60 or better, a life expectancy of at least 8 weeks, and no prior chemotherapy. Only 4/39 (10.2%) patients receiving SMF and 2/36 (5.5%) patients receiving CAC had objective responses. Despite a negligible response rate, the median survival duration was twice as long in the SMF arm than in the CAC arm (10 vs 5 months, P 0.008). In contrast to these favorable results, several other randomized trials failed to substantiate a benefit for the streptozocin combinations; response rates were marginal, ranging from 7% to 12%, and there was no difference in survival. 17,18,26-28 These trials compared versus streptozotocin (median survival, 18.4 vs 16.3 weeks), 26 streptozotocin versus cyclophosphamide

6 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease (median survival, 13 vs 9 weeks), 27 versus streptozotocin tubercidin (median survival not reported), 28 MeCCNU versus MeCCNU streptozotocin (median survival, 14 vs 12 weeks), 18 and investigational agents versus doxorubicin mitomycin C streptozotocin (FAM-S) (median survival, 3.4 vs 4.8 months). 17 Finally, closing the loop around these mitomycin C combinations, the SMF regimen was found to be equivalent to the FAM regimen, which, in turn, had been shown to prolong survival versus best supportive care without, however, improving survival relative to alone. A three-arm study conducted by the Gastrointestinal Tumor Study Group 29 compared the FAM regimen and the SMF regimen (SMF1), both from the Georgetown University group, and a modified SMF regimen (SMF2) that used daily administrations of bolus for 5 days. A total of 130 patients with measurable disease were randomized, including 92 patients with no prior chemotherapy. Response rates for the three arms were virtually identical (13%, 15%, and 14% for FAM, SMF1, and SMF2, respectively), and the corresponding median survival times were 11.6, 17.7, and 13.3 weeks, respectively. Results were very similar when the analysis was restricted to patients with no prior chemotherapy. Severe leukopenia and stomatitis were more common with FAM, whereas emesis was more frequent with the SMF regimens. The Cancer and Leukemia Group B also compared the FAM and the SMF regimens from Georgetown in a randomized trial involving 196 patients. 30 As in the previous study, there was no significant difference between FAM and SMF in terms of response rate (14% vs 4%) and overall survival (median survival, 26.4 vs 18.3 weeks). Of note, the FEM regimen, substituting epirubicin for adriamycin, was evaluated versus epirubicin alone in a randomized trial involving 65 patients. 31 A survival advantage was suggested for the combination over singleagent therapy with a 1-year survival rate of 23.2% versus 15.4%, respectively. That difference was not significant, but the study was not powered for definitive statistical conclusions. Platinum agents have not been extensively evaluated in randomized trials with in pancreatic cancer. Three trials are reported here, and none showed an advantage in survival versus single-agent therapy. Cisplatin was combined with 1,000 mg/m 2 /day given by continuous infusion over 5 days in a comparative trial in which the control was 500 mg/m 2 /day given by bolus injection for 5 days. 32 The role of cisplatin in that trial is difficult to assess considering that each arm used a different method of drug administration for. Among patients with measurable disease, 10 of 98 (10%) experienced an objective tumor regression in the cisplatin arm versus 0 of 98 in the alone arm. Progressionfree survival also favored the combination (median progression-free survival, 73 vs 59 days, P ), but there was no difference in survival (Table 32.2). Of note, a screening randomized phase II trial in 36 patients failed to detect any suggested advantage of -interferon added to a combination of cisplatin and continuous infusion of. 33 The following study was a randomized comparison of a combination of plus doxorubicin plus cisplatin (FAP), the Mallison regimen, and alone. 34 There was no significant difference among the three treatments with respect to survival (Table 32.2). As was the case with the FAM regimen, the Mallison regimen, which had demonstrated superiority over best supportive care in a previous study, 4 was ultimately found to have no advantage over single-agent. A possible role of oxaliplatin was suggested in a preliminary report from a randomized phase II trial that enrolled 65 patients with advanced cancer of the pancreas. 35 Patients were randomized to treatment with single-agent oxaliplatin 130 mg/m 2, single-agent 1 g/m 2 /day for 4 days, or a combination of these drugs at the same doses. All treatments were administered every 3 weeks. There were no responses in either of the single-agent arms and three responses (11%) in the combination arm. The median survival was longer in the combination arm (8.5 months compared with 3.4 and 2.4 months for single-agent oxaliplatin and single-agent, respectively). These data would warrant further investigations if the results are confirmed in the final analysis. Octreotide, a somatostatin analogue, yielded negative results versus and plus leucovorin. 36 In fact, time to progression was significantly shorter for octreotide (42 vs 105 days with chemotherapy, P 0.01). There was, however, no difference in survival, possibly as the result of crossover therapy in octreotide failures. The data were in line with results of other randomized trials of somatostatin analogues in pancreatic cancer indicating a lack of activity when given as single agents versus untreated controls or when given in combination with versus alone. 36 The trial also compared intravenous bolus 500 mg/m 2 for 5 consecutive days and 425 mg/m 2 plus leucovorin 20 mg/m 2 daily for 5 consecutive days. Chemotherapy cycles were repeated every 5 weeks. No meaningful data were given for the comparison of these two treatment options.

7 21786_Ch32.qxd_ /20/05 9:09 AM Page 453 Randomized, Single-Agent Trials 453 Table 32.3 Author(s) Burris et al. 37 Moore et al. 38 Tempero et al. 39 Smith and Gallagher 40 Lersch et al. 41 Abbreviation:, not reported. Single-Agent Randomized Trials Treatment BAY minute infusion 10 mg/m 2 /min ZD9331 SCH No. of Patients (Total) Median Survival 5.65 months 4.41 months 6.4 months 3.2 months 5 months 8 months 109 days 152 days 4.4 months 3.3 months P Value Randomized Trial of Versus In the mid-1990s, gemcitabine replaced as the reference standard treatment in randomized trials for patients with advanced pancreatic cancer. This occurred after a randomized comparison of gemcitabine to in patients with a pathologic diagnosis of locally advanced or metastatic pancreatic cancer and no prior chemotherapy. 37 In that study, a minimum Karnofsky performance status and an estimated life expectancy of at least 12 weeks were required. Patients were assigned to treatment with gemcitabine 1,000 mg/m 2 once weekly for 7 weeks and then, after a 1-week rest period, weekly for 3 out of 4 weeks, or to treatment with bolus 600 mg/m 2 weekly. Unlike previous trials in this setting, the primary efficacy endpoint was a measure of symptom improvement and palliation. The concept of clinical benefit response was developed as a method for assessing the effect of chemotherapy. Clinical benefit response was based on a composite of measurements of pain (analgesic consumption and pain intensity), Karnofsky performance status, and weight. Other efficacy endpoints in the study were response rate, survival, and time to progressive disease. The trial accrued 63 patients in each arm. More than two thirds of the patients in each arm had a performance status of 50 to 70. The number of patients experiencing a clinical benefit response was significantly greater in the gemcitabine group than in the group (15 of 63 [23.8%] vs 3 of 63 [4.8%], P ). Only few objective tumor responses were seen, but there was a statistically significant advantage for gemcitabine in terms of time to progression and survival. Among patients with bidi- mensionally measurable disease at study entry, 3 of 56 (5.4%) achieved a partial response in the gemcitabine arm as compared with 0 of 57 in the arm. Time to progressionfree survival for gemcitabine was 9 weeks compared with 4 weeks for the arm (P ), and the corresponding median survival times were 5.65 months compared with 4.41 months (P ) (Table 32.3). Both treatments were generally well tolerated throughout the study, but there was more myelosuppression with gemcitabine then with, especially in terms of grades 3 4 neutropenia (25.9% vs 4.9%) and grades 3 4 anemia (9.7% vs 0%). The gemcitabine arm was also associated with more fever (30.1% vs 16.1%), more rashes (23.8% vs 12.9%), and more severe emesis (12.7% vs 4.8%) than the arm. Randomized, Single-Agent Trials Randomized, single-agent gemcitabine trials were mostly designed to evaluate the relative role of new investigational agents while attempts at optimizing dose scheduling of gemcitabine were ongoing. The impact on survival of gemcitabine was confirmed in another randomized trial where single-agent gemcitabine was compared with the orally administered matrix metalloproteinase inhibitor, BAY (see Table 32.3). 38 That trial was actually closed early after a planned interim analysis revealed a significant difference in overall survival between treatment groups. At the time of the analysis, a total of 277 patients had been accrued and 140 deaths had occurred. The median overall survival was 6.4 months for the gemcitabine

8 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease arm and only 3.2 months for the BAY arm (P ). Progression-free survival also was longer in the gemcitabine arm, with corresponding median times of 3.5 months versus 1.7 month, respectively (P 0.012). Attempts at further optimizing dose scheduling of gemcitabine are ongoing. A randomized comparison of two gemcitabine regimens in 92 patients (1,500 mg/m 2 given intravenously at 10 mg/m 2 per minute versus a 30-minute intravenous infusion of 2,200 mg/m 2 ) appeared to favor the more prolonged infusion, with a median survival of 8 versus 5 months (p 0.013). 39 These observations were paralleled by greater hematologic toxicity and higher gemcitabine triphosphate levels in mononuclear cells. ZD9331, an intravenously administered thymidylate synthase inhibitor, was compared with gemcitabine in patients with locally advanced or metastatic pancreatic cancer. 40 The study was stopped prematurely because of two drug-related deaths that occurred in the ZD9331 arm after the first cycle of treatment. Full assessment of the data available from 30 patients randomized to ZD9331 and 25 patients randomized to gemcitabine before the time the study was terminated showed median survival, 152 versus 109 days, and time to progression, 70 versus 58 days, favoring ZD9331. Tumor response and clinical benefit response rates were similar for both groups. More patients in the ZD9331 treatment group discontinued treatment because of toxicity, primarily myelosuppression, than did patients treated with gemcitabine. Gastrointestinal toxicities, including grade 3 nausea, vomiting, and diarrhea, were more common with ZD9331 treatment. The apparent improvement in median survival and time to progression observed with ZD9331 should be considered in the context of the toxicity associated with this treatment. In a randomized phase II study, previously treated and chemotherapy-naïve patients were allocated to receive treatment with gemcitabine (n 30) or SCH (n 33), an orally administered farnesyl transferase inhibitor. 41 There was one partial response in the gemcitabine group, and there were two partial responses in the SCH group. The 3- month progression-free survival rate, the primary endpoint of the trial, favored the gemcitabine arm over SCH (31% vs 23%) and so were the median overall survival times (4.4 vs 3.3 months). Randomized Trials of - Based Combination Therapy Several gemcitabine combinations were evaluated in randomized trials versus single-agent gemcitabine therapy (Table 32.4). Similar to what was observed when single- agent was compared with containing combination regimens, tumor response rates remained generally low. Some superiority in response rates was reported with cisplatin and irinotecan in combination with gemcitabine, and time to progression was improved with and cisplatin combinations. However, with a single exception (NSC ), this series of studies did not identify a combination regimen that prolonged survival relative to gemcitabine alone. The Eastern Oncology Cooperative Group (ECOG) compared gemcitabine 1000 mg/m 2 plus bolus 600 mg/m 2 to the same dose of gemcitabine alone. 42 All drugs were given weekly for 3 of 4 weeks. Eligibility criteria included microscopically confirmed measurable or assessable disease and an ECOG performance status of 0 2. Of 327 patients enrolled, 322 were eligible. There were no complete responses in either of the treatment groups. The partial response rates were 6.9% and 5.6% for patients receiving gemcitabine plus and single-agent gemcitabine, respectively. There was a statistically significant difference in progression-free survival favoring gemcitabine plus (3.4 vs 2.2 months, P 0.022). This difference, however, did not translate into a difference in survival, the primary endpoint for the study. The median overall survival was 6.7 months for the combination regimen and 5.4 months for gemcitabine alone. That difference did not reach the 0.05 level of significance, unless adjusted for an imbalance in baseline performance status (P 0.037). The regression analysis suggested that survival results could have been different had patients been stratified by performance status, but this observation was felt unimportant in view of the minimal survival gain. Toxicities were also more common in the combination treatment group. by continuous infusion in combination with gemcitabine did not result in any difference in response rate, progression-free, or overall survival as compared to gemcitabine alone. 43 The response rate for the combination was 13%, and it was 9% with single-agent gemcitabine. The median progression-free survival was 3 months in both arms, and the median overall survival was also the same in both arms at 6 months. There was somewhat more grade 3 to grade 4 toxicity with the combination than with single-agent therapy. Three randomized phase III studies evaluated a combination of gemcitabine plus cisplatin compared with gemcitabine alone. The two largest trials showed a benefit in terms of time to progression, but not in terms of survival. These trials may have been inadequately powered to detect a small difference in survival. A trial conducted in Italy used single-agent gemcitabine at a dose of 1000 mg/m 2 weekly for 7 consecutive weeks

9 21786_Ch32.qxd_ /20/05 9:09 AM Page 455 Randomized Trials of -Based Combination Therapy 455 Table 32.4 Randomized Trials of -Based Combination Regimens Versus Alone Author(s) Treatment No. of Patients Median Survival 6.7 months 5.4 months 6.0 months 6.0 months 30 weeks 20 weeks 8.3 months 6.0 months 217 days 273 days 9.5 months 8.2 months 6.3 months 6.6 months days 164 days 193 days 182 days 10.4 months 7.9 months 5.2 months 6.2 months 6.3 months 9.0 months 7.1 months P Value Berlin et al. 42, Di Costanzo et al. 43, (CI) Colucci et al. 44, Cisplatin Heinemann et al. 45, Cisplatin Wang et al. 46, Cisplatin Scheithauer et al. 47, Capecitabine Rocha Lima et al. 48, Irinotecan Bramhall et al. 50, Marimastat Van Cutsem et al. 51, tipifarnib Gansauge et al. 53, NSC NSC Richards et al. 54, permetrexed Louvet et al. 55, oxaliplatin Abbreviations:, not reported; NS, not significant NS followed, after a 2-week rest, by drug administration on days 1, 8, and 15 of a 28-day cycle for two cycles. 44 In the combination arm, cisplatin was given at a weekly dose of 25 mg/m 2, and gemcitabine was given as in the single-agent arm. A total of 107 patients were entered; 53 were randomized to receive the gemcitabine plus cisplatin combination, and 54 were randomized to receive gemcitabine alone. That sample size had been selected to demonstrate a 30% improvement in time to progression. The overall response rate for the gemcitabine plus cisplatin arm was 26.4% compared with 9.2% for gemcitabine alone (P 0.02). The median time to progression was 20 weeks for the combination and 8 weeks for single-agent gemcitabine (P 0.048), whereas the median overall survival was 30 weeks for the combination arm and 20 weeks for single-agent gemcitabine arm (P 0.43). There was no difference in clinical benefit derived from a composite assessment of pain, functional impairment, and weight loss. Toxicity was reportedly mild in both arms, but asthenia was more frequent with the combination. Preliminary results from a larger study conducted in Germany also indicated a significant advantage in progression-free survival, but no significant superiority in terms of survival. 45 A total of 195 patients were enrolled. A combination of gemcitabine 1,000 mg/m 2 and cisplatin 50 mg/m 2 on days 1 and 15 of a 28-day cycle was compared with gemcitabine alone for 1000 mg/m 2 on days 1, 8, and 15 of a 28-day cycle. Progression-free survival markedly favored the combination over the single agent (5.4 vs 2.8 months, P 0.01), but the advantage in survival (8.3 vs 6.0 months) was not significant (P 0.12). Except for nausea and vomiting, 20.9% in the combination arm, and 6.4% in the single-agent arm, toxic effects were similar in both treatment groups.

10 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease Wang et al. 46 randomized 22 patients to treatment with a combination of gemcitabine plus cisplatin and 20 patients to treatment with gemcitabine alone. That small study did not show any difference in response rate or survival. The oral prodrug, capecitabine (1,250 mg/m 2 twice a day for 7 days), was given in combination with gemcitabine (2,200 mg/m 2 every 2 weeks) and compared with the same dose and schedule of gemcitabine alone in a randomized phase II trial. 47 Sample sizes were predetermined to detect an increase by 50% in time to progression at 5 months, relative to historical control subjects. A total of 83 patients were entered. The combination and single-agent treatment groups yielded similar results with respect to objective response (17% vs 14%), progressionfree survival (5.1 vs 4.0 months), overall survival (9.5 vs 8.2 months), and clinical benefit response (48.4% vs 33%). Adverse reactions were more severe with the combination, especially with the addition of the capecitabine-related hand foot syndrome, but dose reductions, treatment delays, or treatment discontinuations were fairly similar with both treatment options. Irinotecan in combination with gemcitabine produced significantly better tumor response rates than gemcitabine alone, but this combination also failed to prolong significantly the time to progression or survival. 48 A total of 360 chemotherapy-naïve patients with locally advanced or metastatic pancreatic cancer were randomized after stratification by performance status, extent of disease, and prior radiotherapy to receive either gemcitabine 1,000 mg/m 2 immediately followed by irinotecan 100 mg/m 2 on days 1 and 8 of 3-week cycles or gemcitabine alone at a dose of 1,000 mg/m 2 weekly for 7 out of 8 consecutive weeks on cycle 1 and then 1,000 mg/m 2 on days 1, 8, and 15 of 4-week cycles. In a preliminary analysis, the overall response rate was 16.1% versus only 4.4% (P 0.001). The median time to progression was 3.4 versus 3.0 months, and the median survival time was 6.3 versus 6.6 months. Combination chemotherapy was associated with more severe diarrhea than single-agent chemotherapy (18.5% vs. 1.8%). Docetaxel has also been combined with gemcitabine. That combination was tested versus a combination of docetaxel and cisplatin in a randomized phase II trial that involved 96 patients. 49 No difference was found in terms of response rate, progression-free survival, or survival. was used in randomized trials in combination with a number of investigational agents, including marimastat, 50 tipifarnib, 51 and CI (Table 32.4). In these three cases, a double-blind design was used to compare the two-drug regimens to gemcitabine and placebo. Adding marimastat, a matrix metalloproteinase inhibitor, did not result in any benefit in terms of response rate, progression-free survival, or overall survival, based on a trial that accrued 239 patients. The combination of tipifarnib, a farnesyltransferase inhibitor, with gemcitabine also failed to show an advantage in survival as compared with gemcitabine alone. There was, however, more severe neutropenia and more diarrhea with the combination. No published data are available, as yet, on the randomized phase II trial of CI-994, an oral histone deacetylase inhibitor, but the study was reported to show no difference in survival rates and perhaps an increase in toxicities. 52 NSC (Ukrain) is a preparation derived from the Greater Celandine, a common herb, and thiotepa. It has been used in alternative medicine as an anticancer drug. It was tested alone or in combination with gemcitabine versus single-agent gemcitabine therapy in a three-arm phase II study performed at the University of Ulm in Germany. 53 A total of 90 patients were randomized. Limited details are available regarding study design, study execution, or statistical considerations. A significant survival advantage was reported with both investigational arms relative to gemcitabine alone (P 0.01 in both comparisons), with no significant increase in toxicity. These findings were unexpected and make NSC worthy of further investigation. Recent reports support a superiority of gemcitabine combined with permetrexed 54 or oxaliplatin 55 over gemcitabine alone in terms of response rate 54,55, and time to progression 54 or progression-free survival. 55 These combinations, however, failed to achieve a significant difference in survival. Both were associated with significantly more grade 3 and 4 toxicities. The lithium salt of gamolenic acid, a highly unsaturated fatty acid, was also evaluated in a randomized trial that compared oral and intravenous administration at two doses. Results were comparable in all treatment groups. There was no known active control in that trial, but median survival times were reminiscent of historical controls treated with best supportive care. 56 Randomized Second-Line Therapy Trials The increased use of gemcitabine in first-line chemotherapy has now generated a new need for second-line treatment. Raltitrexed, a thymidilate synthase inhibitor, was evaluated in a randomized study as a single agent or in combination with irinotecan. 57 The trial accrued 38 patients with metastatic pancreatic adenocarcinoma, whose disease had progressed while receiving or within 6 months after discontinuation of palliative first-line chemotherapy with gemcitabine. The primary endpoint was response rate, and the study was terminated early because of a lack of tumor response in the single-agent raltitrexed arm (0%) compared

11 21786_Ch32.qxd_ /20/05 9:09 AM Page 457 Future Directions 457 Table 32.5 Ongoing Randomized Trials in Advanced/Metastatic Pancreatic Cancer Phase Investigator (Group or Location) Treatment Phase II Eastern Cooperative Oncology Group Arm A: Irinotecan + Docetaxel + Cetuximab Arm B: Irinotecan + Docetaxel Phase II North Central Cancer Treatment Group Arm A: Bortezomib Arm B: + Bortezomib Phase II Cancer and Leukemia Group B Arm A: Arm B: + Cisplatin Arm C: + Docetaxel Arm D: + Irinotecan Phase II Fox Chase Arm A: Bevacizumab + Docetaxel Arm B: Bevacizumab Phase III Eastern Cooperative Oncology Group Arm A: 30-minute infusion Arm B: 150-minute infusion Arm C: 100-minute infusion + Oxaliplatin Phase III Swiss Institute for Applied Cancer Research/Central Arm A: + Capecitabine European Cooperative Oncology Group Arm B: Phase III Cancer Research UK Arm A: + Capecitabine Arm B: Phase III Wellstat Therapeutics Arm A: + Oral Triacetyl Uridine Arm B: Phase III Lorus Therapeutics Arm A: + Virulizin Arm B: Phase III Southwest Oncology Group Arm A: 30 minute infusion + Ce tuximab Arm B: 30 minute infusion Phase III Cancer and Acute Leukemia Group B Arm A: 30 minute infusion + Bevacizumab Arm B: 30 minute infusion Arranged from National Cancer Institute Clinical Trials. 58 with a 16% response in the combination arm. The clinical benefit response rate also favored the combination arm (29% vs 8%), as did median progression-free survival (4.0 vs 2.5 months) and median survival (6.5 vs 4.3 months). Future Directions Despite over 20 years of clinical investigations, the goal of producing meaningful prolongation of survival for patients with advanced pancreatic cancer remains elusive, and the prognosis for these patients remains extremely poor. Only a few agents have demonstrated anti-tumor activity in pancreatic cancer, including primarily and gemcitabine. There are no randomized trials establishing an impact on survival with single-agent therapy in this disease, but combinations have been shown to prolong survival as compared with best supportive care, and comparisons of several combinations versus alone have failed to demonstrate prolonged survival relative to alone. is another active agent that impacts on survival, as demonstrated in trials versus and versus a matrix metalloproteinase inhibitor. A superiority over has, however, been questioned, as both and gemcitabine were possibly not compared at optimal dose schedules. The combination of and gemcitabine has shown some benefit, but the advantage in survival versus gemcitabine alone was not pronounced enough to achieve strong statistical significance. Some agents, including flutamide and Ukrain, have unexpectedly been reported to prolong survival in relatively smallsize randomized trials and confirmatory trials are needed for more definitive answers. Elucidating molecular processes in cancer cells has identified a number of attractive targets. Although initial results from randomized studies using therapies targeted against matrix metalloproteinase, farnesyl transferase, or histone deacetylase have been disappointing, targeted therapies remain an active area of investigation (Table 32.5). 58 Ongoing randomized phase II trials are evaluating treatment options other than gemcitabine in chemotherapy-

12 21786_Ch32.qxd_ /20/05 9:09 AM Page Advanced Metastatic Disease naïve patients with metastatic disease, including cetuximab, a monoclonal antibody that targets the epidermal growth factor receptor, and bortezomib, a proteasome inhibitor. The former has been incorporated in a regimen combining irinotecan and docetaxel. The latter is used as singleagent therapy and in combination with gemcitabine. Some phase II trials are also open to patients with prior exposure to chemotherapy. A number of two-drug combinations of gemcitabine with cisplatin, docetaxel, or irinotecan are being tested versus gemcitabine alone in patients who may have received prior. Patients with prior exposure to gemcitabine are eligible for a trial of bevacizumab, a humanized monoclonal antibody directed against vascular endothelial growth factor. All of the reported ongoing phase III trials use singleagent therapy with gemcitabine as a control. The issues that are being currently addressed concern the optimization of gemcitabine administration or the enhancement of the therapeutic index of single-agent gemcitabine with the addition of other agents. The role of capecitabine, oxaliplatin, irinotecan, and docetaxel is being further elucidated. Identifying new active agents from new classes of anti-cancer agents remains a high priority area for research in this disease. References 1. Jemal A, Tiwari RC, Murray T, et al. Cancer Statistics, CA Cancer J Clin. 2004;54: El Kamar FG, Grossbard ML, Kozuch PS. Metastatic pancreatic cancer: emerging strategies in chemotherapy and palliative care. Oncologist. 2003;8: Abbruzzese JL. Past and present treatment of pancreatic adenocarcinoma: chemotherapy as a standard treatment modality. Semin Oncol. 2002;29(Suppl 20):S2 S8. 4. Haller DG. New perspectives in the management of pancreas cancer. Semin Oncol. 2003;30(Suppl 11):S3 S Mallinson CN, Rake MO, Cocking JB, et al. Chemotherapy in pancreatic cancer: results of a controlled, prospective, randomised, multicentre trial. Br Med J. 1980;281: Palmer KR, Kerr M, Knowles G, et al. Chemotherapy prolongs survival in inoperable pancreatic carcinoma. Br J Surg. 1994;81: Glimelius B, Hoffman K, Sjödén P-O, et al. Chemotherapy improves survival and quality of life in advanced pancreatic and biliary cancer. Ann Oncol. 1996;7: Andersen JR, Friis-Møller A, Hancke S, et al. A controlled trial of combination chemotherapy with and BCNU in pancreatic cancer. Scand J Gastroenterol. 1981;16: Frey C, Twomey P, Keehn R, et al. Randomized study of and CCNU in pancreatic cancer: report of the Veterans Administration Surgical Adjuvant Cancer Chemotherapy Study Group. Cancer. 1981;47: Andren-Sandberg A, Holmberg JT, Ihse I. Treatment of unresectable pancreatic carcinoma with 5-fluorouracil, vincristine and CCNU. Scand J Gastroenterol. 1983;18: Bakkevold KE, Pettersen A, Arnesjo B, Espehaug B. Tamoxifen therapy in unresectable adenocarcinoma of the pancreas and the papilla of Vater. Br J Surg. 1990;77: Keating JJ, Johnson PJ, Cochrane AMG, et al. A prospective randomised controlled trial of tamoxifen and cyproterone acetate in pancreatic carcinoma. Br J Cancer. 1989;60: Greenway BA. Effect of flutamide on survival in patients with pancreatic cancer: results of a prospective, randomised, double blind, placebo controlled trial. BMJ. 1998;316: Schein PS, Lavin PT, Moertel CG, et al. Randomized phase II clinical trial of adriamycin, methotrexate and actinomycin-d in advanced measurable pancreatic carcinoma: a Gastrointestinal Tumor Study Group Report. Cancer. 1978;42: Gastrointestinal Tumor Study Group. Phase II trials of hexamethylmelamine, dianhydrogalactitol, razoxane, and -2 deoxythioguanosine as single agents against advanced measurable tumors of the pancreas. Cancer Treat Rep. 1985;69: Gastrointestinal Tumor Study Group. Phase II trials of maytansine, low-dose chlorozotocin, and high-dose chlorozotocin as single agents against advanced measurable adenocarcinoma of the pancreas. Cancer Treat Rep. 1985;69: Bukowski RM, Fleming TR, Macdonald JS, et al. Evaluation of combination chemotherapy and phase II agents in pancreatic adenocarcinoma. A Southwest Oncology Group Study. Cancer. 1993;71: Horton J, Gelber RD, Engstrom P, et al. Trials of single-agent and combination chemotherapy for advanced cancer of the pancreas. Cancer Treat Rep. 1981;65: Kovach JS, Moertel CG, Schutt AJ, et al. A controlled study of combined 1,3-BIS-(2-chloroethyl)-1-nitrosourea and 5-fluorouracil therapy for advanced gastric and pancreatic cancer. Cancer. 1974;33: Stolinsky DC, Pugh RP, Bateman JR. 5-Fluorouracil (NSC ) therapy for pancreatic carcinoma: comparison of oral and intravenous routes. Cancer Chemother Rep. 1975;59: Maisey N, Chau I, Cunningham D, et al. Multicenter randomized phase III trial comparing protracted venous infusion (PVI) fluorouracil () with PVI plus mitomycin in inoperable pancreatic cancer. J Clin Oncol. 2002;20: Cullinan SA, Moertel CG, Fleming TR, et al. A comparison of three chemotherapeutic regimens in the treatment of advanced pancreatic and gastric carcinoma. Fluorouracil vs fluorouracil and doxorubicin vs fluorouracil, doxorubicin, and mitomycin. JAMA. 1985;253: Takada T, Kato H, Matsushiro T, et al. Comparison of 5-fluorouracil, doxorubicin and mitomycin C with 5-fluorouracil alone in the treatment of pancreatic-biliary carcinomas. Oncology. 1994;51: Bukowski RM, Balcerzak SP, O Bryan RM, et al. Randomized trial of 5-fluorouracil and mitomycin C with or without streptozotocin for advanced pancreatic cancer: a Southwest Oncology Group Study. Cancer. 1983;52: